Comparison of whole-body DW-MRI with 2-[18F]FDG PET for staging and treatment monitoring of children with Langerhans cell histiocytosis

Abstract

Purpose: To assess and compare the diagnostic accuracy of whole-body (WB) DW-MRI with 2-[¹⁸F]FDG PET for staging and treatment monitoring of children with Langerhans cell histiocytosis (LCH).

Methods: Twenty-three children with LCH underwent 2-[¹⁸F]FDG PET and WB DW-MRI at baseline. Two nuclear medicine physicians and two radiologists independently assessed presence/absence of tumors in 8 anatomical areas. Sixteen children also performed 2-[¹⁸F]FDG PET and WB DW-MRI at follow-up. One radiologist and one nuclear medicine physician revised follow-up scans and collected changes in tumor apparent diffusion (ADC) and standardized uptake values (SUV) before and after therapy in all detectable lesions. 2-[¹⁸F]FDG PET results were considered the standard of reference for tumor detection and evaluation of treatment response according to Lugano criteria. Sensitivity, specificity, positive and negative predictive values, and diagnostic accuracy of WB DW-MRI at baseline were calculated, and the 95% confidence intervals were estimated by using the Clopper-Pearson (exact) method; changes in tumor SUVs and ADC were compared using a Mann-Whitney U test. Agreement between reviewers was assessed with a Cohen's weighted kappa coefficient. Analyses were conducted using SAS software version 9.4.

Results: Agreement between reviewers was perfect (kappa coefficient = 1) for all analyzed regions but spine and neck (kappa coefficient = 0.89 and 0.83, respectively) for 2-[¹⁸F]FDG PET images, and abdomen and pelvis (kappa coefficient = 0.65 and 0.88, respectively) for WB DW-MRI. Sensitivity and specificity were 95.5% and 100% for WB DW-MRI compared to 2-[¹⁸F]FDG PET. Pre to post-treatment changes in SUV_ratio and ADC_mean were inversely correlated for all lesions (r: -0.27, p = 0·06) and significantly different between responders and non-responders to chemotherapy (p = 0.0006 and p = 0·003 for SUV_ratio and ADC_mean, respectively).

Conclusion: Our study showed that WB DW-MRI has similar accuracy to 2-[¹⁸F]FDG PET for staging and treatment monitoring of LCH in children. While 2-[¹⁸F]FDG PET remains an approved radiological examination for assessing metabolically active disease, WB DW-MRI could be considered as an alternative approach without radiation exposure. The combination of both modalities might have advantages over either approach alone.

Keywords: 2-[18F]FDG PET/MRI; Pediatric; Staging; Treatment monitoring; WB DW-MRI, LCH.

Fig. 1

Flowchart. *Mean ± standard deviation; **girl/boy; CHOP, Children’s Hospital of Philadelphia; LCH, Langerhans cell histiocytosis

Fig. 2

Discordant finding on baseline 2-[¹⁸F]FDG PET and DW-MRI in a 2-year-old girl. (A) Axial 2-[¹⁸F]FDG PET, (B) T1-weighted gradient echo (LAVA) MRI, (C) fused 2-[¹⁸F]FDG PET/LAVA MRI show a metabolically active lesion in the cervical spine (arrow). (D) Axial DW-MRI, (E) LAVA MRI, (F) fused DW-MRI/LAVA MRI. Either axial DW-MRI or fused DW-MRI/LAVA MRI demonstrate no abnormal finding in the cervical spine

Fig. 3

(a) Concordant baseline findings on 2-[¹⁸F]FDG PET and DW-MRI in a 12-year-old boy. At baseline (A) axial 2-[¹⁸F]FDG PET and (B) fused 2-[¹⁸F]FDG PET/T1-weighted gradient echo (LAVA) MRI as well as (C) axial DW-MRI and (D) fused DW-MRI/LAVA MRI show a metabolically active and diffusion restricted bone lesion in the right ilium, respectively. (b) Concordant response to treatment on 2-[¹⁸F]FDG PET and DW-MRI in a 12-year-old boy. On post-treatment imaging (A) axial 2-[¹⁸F]FDG PET and (B) fused 2-[¹⁸F]FDG PET/T1-weighted gradient echo (LAVA) MRI as well as (C) axial DW-MRI and (D) fused DW-MRI/LAVA MRI demonstrate a partial response with reduced 2-[¹⁸F]FDG metabolism and less diffusion restriction, respectively

Fig. 4

Concordant nonresponse to treatment on 2-[¹⁸F]FDG PET and DW-MRI in a 2-year-old male. (A–D) At baseline (A) axial 2-[¹⁸F]FDG PET and (B) fused 2-[¹⁸F]FDG PET/T1-weighted gradient echo (LAVA) MRI as well as (C) axial DW-MRI and (D) fused DW-MRI/LAVA MRI demonstrate no lesion in the mid-thigh. (E–H) On post-treatment imaging after 4 months from baseline (E) axial 2-[¹⁸F]FDG PET and (F) fused 2-[¹⁸F]FDG PET/LAVA MRI show two new lesions in the vastus lateralis muscle with increased 2-[.¹⁸F]FDG uptake (arrows); similarly, (G) axial DW-MRI and (H) fused DW-MRI/LAVA MRI demonstrate restricted lesions in thigh vastus lateralis muscles (arrows)

Fig. 5

SUV_ratio and ADC_mean change by response status at baseline and post-treatment. The figure shows box and whisker plots which compare (A and B) the SUV_ratio and ADC_mean values between baseline and follow-up for responders and non-responders to chemotherapy. Resolution of the lesions leads to a decrease in SUV_ratio and an increase in ADC_mean values. (C and D) The change in SUV_ratio and ADC_mean between baseline and follow-up scan was significantly different between responders and non-responders (Mann-Whitney U test p = 0.0006 and p = 0.003, respectively)